Today I got my Kinect V2 in an attempt to make a budget 3d scanner it didn't work out too well but thought I would still make a video encase anyone else is considering doing the same.
Both of these programs use the background as reference, and it recommended to leave the object stationary and moving the camera. This explains your scans
OH MY GOD! YOU ARE DOING COMPLETELY WRONG! The kinect NEEDS background movement to WORK when 3d scanning, with the kinect, you CAN'T just put something on a turntable and expect it to work. this video is SO TRIGGERING lol I was cringing at 2:56 because the object is BLACK! that means that you need to MOVE the kinect back from the object!!!!
KScan3D was the best option I found for Kinect v2, got decent enough results when I tried with a lot of export options. Worth a try for sure but Kinect probably can't compete with photogrammetry at this point even though the tech is far more capable, there just isn't the level of support it requires. I tested a few of the Kinect v1 options too and they were way more advanced. Had Kinect v2 been embraced like v1 was there could've been some wildly impressive software written for it.
I remember seeing people who made those expensive robots that can lift thing's and jump up stairs etc, and the one thing that stuck out to me was the robot used 2 separate Kinects combined to work together to give the robot it's spacial awareness/ infrared camera/ and I think the other kinect was used to always be scanning everything in the room to guide the robot and show depth perception. Always thought the kinect had so much potential but once MS stopped supporting and selling the V2 especially since it hadn't been out nearly as long as the V1 kinect had been when they stopped supporting it....because I remember some of the last 360 consoles they made (most recent) Xbox 360 slim models that had all had that specific orange port just for the Kinect and they were selling Kinect V1 bundled with some or most of the last 360s they sold. Anyways, my point to all that rambling was if the V2 Kinect had been actively used and shipped with the later versions of consoles after the first gen Xbox One (Xbox One S/Xbox One X) then we would have much more updates, PC drivers, and support like we did for the nearly 10 years we had the V1 Kinect for
You have to use the depth to cut off your background while scanning. You are just moving an object in front of a static background. The kinect notices the Background. The Background is always the same, that is why your moving parts get shitty results. Try to use a background that ist 3 to 4 meters away. The result will be really good. But the object you scan has to be static. you have to rotate the object, but don't move your head reativly to your body. Good luck. I got some good results.
my first go on 3Dscan with windows yielded much better results than you are showing here, and consistently the results ive been getting have been from OK to very good. not sure what issue you are having here... I think that either you may be too close to the camera ( I know for a fact Kinects depth sensing craps out 30 - 40 cm from the sensor ) or you had poor lighting. probably a combination of the two - Im also using Kinect to drive pointclouds and 3d meshes in real time with touchdesigner, so while I understand your frustration at the device seemingly not working as intended, I think your portrayal of the kinects functionality in this regard is off the mark
i used, and it takes some tries, but it does work i placed mine above me and pointed it down, use the auto start and auto stop.. and ensure when i turned byself i did not move my head, and DONT do double scan of yourself like scaniverse on iphone with a lidar sensor, think of the camera as a spray paint and dont spray paint yourself twice.. for me the auto end at 10 sec seems to work the best for me
I'm in the same boat here. Skannect I bought some years ago and have gotten plenty of incredible results, but the color texture resolution is rough. The V2 sadly doesn't work with Skannect for whatever reason. 3D scan I can't even get to read my V2, while everything else does. The V2 is still a fantastic mocap tool. A shame support for a greatly improved product is nonexistent for scanning.
I tried that too it really didnt give me anything usuable. I did some research and the reson that scanekt doesnt support the kinect v2 is because they couldnt get usuable results
This video is 2 years old. But dude. You are using it completely wrong in every way possible. You move objects while the kinect is stationary, your ment to do it the other way around. and you constantly have objects way too close, indicated by the black areas. And after using it completely wrong you have the audacity to say it sucks. Even tho the only failure point is your in this process
@@BenitoFan760 yeap.. havent used it for a while.. so going to get it runing again.. but have scanned myself in etc.. most important thing was pc spec.. i was using gpu processing.. so maybe thats what this guys problem is.
@@tsandu I can't get it to recognize the sensor on my windows 10 pc so I'm doing a fresh windows 8 install right now 😂 I have an i5 3470 quad core and a gtx 1050 ti so my guess is just that windows 10 is being dumb
@@BenitoFan760 did you install the kinetic windows driver? Is been some time since i used it but dont recall on having any issue.. oh also did you get the cable to be able to connect it to a pc and power it up?
Both of these programs use the background as reference, and it recommended to leave the object stationary and moving the camera. This explains your scans
OH MY GOD! YOU ARE DOING COMPLETELY WRONG! The kinect NEEDS background movement to WORK when 3d scanning, with the kinect, you CAN'T just put something on a turntable and expect it to work.
this video is SO TRIGGERING lol I was cringing at 2:56 because the object is BLACK! that means that you need to MOVE the kinect back from the object!!!!
Your supposed to move the kinect around the object not the object itself and you have to have good lighting as well
No you don't you can put the object on a lazy Susan which is much better than moving the kinect you carrot
KScan3D was the best option I found for Kinect v2, got decent enough results when I tried with a lot of export options. Worth a try for sure but Kinect probably can't compete with photogrammetry at this point even though the tech is far more capable, there just isn't the level of support it requires.
I tested a few of the Kinect v1 options too and they were way more advanced. Had Kinect v2 been embraced like v1 was there could've been some wildly impressive software written for it.
It's alright if you know how to use it. With 3D scan you need to reduce depth, width and height which wasn't shown in this video
I don't understand why the dev community for the inferior version of kinect v1didn't stick around for kinect v2.
@@MrGamelover23 I managed to download 3D scan version 1 that's also the best version to use with Kinect V2
I remember seeing people who made those expensive robots that can lift thing's and jump up stairs etc, and the one thing that stuck out to me was the robot used 2 separate Kinects combined to work together to give the robot it's spacial awareness/ infrared camera/ and I think the other kinect was used to always be scanning everything in the room to guide the robot and show depth perception. Always thought the kinect had so much potential but once MS stopped supporting and selling the V2 especially since it hadn't been out nearly as long as the V1 kinect had been when they stopped supporting it....because I remember some of the last 360 consoles they made (most recent) Xbox 360 slim models that had all had that specific orange port just for the Kinect and they were selling Kinect V1 bundled with some or most of the last 360s they sold.
Anyways, my point to all that rambling was if the V2 Kinect had been actively used and shipped with the later versions of consoles after the first gen Xbox One (Xbox One S/Xbox One X) then we would have much more updates, PC drivers, and support like we did for the nearly 10 years we had the V1 Kinect for
@@ravenjamtay5347 yeah, it seems like the depth of field & the lighting are the problem with these scans, not the hardware or software.
You have to use the depth to cut off your background while scanning. You are just moving an object in front of a static background. The kinect notices the Background. The Background is always the same, that is why your moving parts get shitty results. Try to use a background that ist 3 to 4 meters away. The result will be really good. But the object you scan has to be static. you have to rotate the object, but don't move your head reativly to your body. Good luck. I got some good results.
Good chuckle at the prospect of free software
“Worth a try but don’t bother” is contradictory 😂
You made many mistake in the process, you results are quite bad. You can achieve much better result with the appropriate workflows
You also needed to reduce the depth, width and height or you going to end up with the background
Great info. Kinect is meant for room-scale scanning/tracking so won't do smaller objects close to it very well.
my first go on 3Dscan with windows yielded much better results than you are showing here, and consistently the results ive been getting have been from OK to very good. not sure what issue you are having here... I think that either you may be too close to the camera ( I know for a fact Kinects depth sensing craps out 30 - 40 cm from the sensor ) or you had poor lighting. probably a combination of the two - Im also using Kinect to drive pointclouds and 3d meshes in real time with touchdesigner, so while I understand your frustration at the device seemingly not working as intended, I think your portrayal of the kinects functionality in this regard is off the mark
The black / grey layer means you are too close.
Thank you for explaining this.
The comments are way more helpful than this video. Thanks, folks.
i used, and it takes some tries, but it does work
i placed mine above me and pointed it down, use the auto start and auto stop.. and ensure when i turned byself i did not move my head, and DONT do double scan of yourself
like scaniverse on iphone with a lidar sensor, think of the camera as a spray paint and dont spray paint yourself twice..
for me the auto end at 10 sec seems to work the best for me
With version 2 you need to use 3D scan along with 3D builder. Also did you update your driver through windows settings?
Help me, I don't have file app 3d scan, Do you have for me. Thanks
Dude seriously! Room lighting is horrible and background too. This was not a TRY by any stretch.
User error.
I'm in the same boat here. Skannect I bought some years ago and have gotten plenty of incredible results, but the color texture resolution is rough.
The V2 sadly doesn't work with Skannect for whatever reason. 3D scan I can't even get to read my V2, while everything else does.
The V2 is still a fantastic mocap tool. A shame support for a greatly improved product is nonexistent for scanning.
Your lighting looks off?
Your lighting is not bright enough I use mine in dim room it does the same . Turn on the bright lights and it's amazing
I have kept mine in the basement for over 10 years. I can let my kids play around with it.
I've been trying to figure out if the Xbox One connect will be better than 360 connect
I tried with less light, then more light, then quicker, so it will make less pictures to stitch, no good scan at the end.
What about moving the kinect around the object?
I tried that too it really didnt give me anything usuable. I did some research and the reson that scanekt doesnt support the kinect v2 is because they couldnt get usuable results
Bro you didn’t put any link in the description
after watching a few videos i saw one it seemed microsoft has a program in the pc app store for this that may work better
What is it called please?
I'm just popping by for a like and a comment of how appreciative I am you got straight to the point in this title. Cheers.
It is worth a try I’ve read a university paper comparing it to a real sense and it beat it
This video is 2 years old. But dude. You are using it completely wrong in every way possible. You move objects while the kinect is stationary, your ment to do it the other way around. and you constantly have objects way too close, indicated by the black areas.
And after using it completely wrong you have the audacity to say it sucks. Even tho the only failure point is your in this process
I did follow your steps but the
kinect 2 keeping disconnected
please help how to fix it
you need renesans 701/702 chipset usb 3 or compatible usb hub or else forget it.
interesting experiment
You didn’t set yourself up for success. Bad lighting, noisy background,and improper handling will surely result in an inadequate outcome.
You had the camera too close
there are other programs that work well with the Kinect1
You got too close to the objects
for touchdesigner, just download the same sdk like you and that's it? Windows 11 obvious
Mine works excellent but I have the other xbox camera
You have the XBOX 360 sensor?
@@BenitoFan760 yeap.. havent used it for a while.. so going to get it runing again.. but have scanned myself in etc.. most important thing was pc spec.. i was using gpu processing.. so maybe thats what this guys problem is.
@@tsandu I can't get it to recognize the sensor on my windows 10 pc so I'm doing a fresh windows 8 install right now 😂 I have an i5 3470 quad core and a gtx 1050 ti so my guess is just that windows 10 is being dumb
@@BenitoFan760 did you install the kinetic windows driver? Is been some time since i used it but dont recall on having any issue.. oh also did you get the cable to be able to connect it to a pc and power it up?
@@BenitoFan760 you have to use the like 1.8 or whatever lower version, 2.0 doesn't work with the older kinect
skill issue
Move arround the object.... ur kinect use the background to Check what is the object ect.
Reconstruct me works really well and has unlimited polygons for free